Sleep and Emotional Reactivity in Alcohol Use Disorder

SPECIFIC AIMS Alcohol use disorder (AUD) is a multifaceted, chronic relapsing disorder suffered by 9.8 million men and 5.3 million women in the United States.4 Current treatment approaches have met with only limited success in preventing relapse,5 although pharmacotherapy that improved sleep in AUD was recently shown to help maintain abstinence.6-9 Targeting multidimensional pathways may more effectively treat AUD and prevent relapse than targeting singular pathways. Here, our investigators aim to investigate links between two factors important in AUD and relapse: altered sleep, specifically rapid eye movement (REM) sleep, and poor emotion regulation. Chronic alcohol use is associated with disrupted sleep continuity and structure, which impact health-related quality of life, and contribute to relapse.10 Polysomnographic (PSG) studies show improvements with extended abstinence, however, sleep does not always return to normal levels, and sleep alterations are predictors of relapse.11-13 Our investigators and others have shown that REM sleep is persistently higher in recovering AUD patients relative to healthy controls (CTL), suggesting chronic changes to REM sleep homeostasis,2,10,13,14 which are likely maladaptive. Importantly, excess duration and intensity of REM sleep (high REM pressure) are predictors of relapse.15,16 Attention has mostly focused on sleep continuity and slow wave sleep measures in AUD recovery. The functional significance of excess REM sleep in AUD is poorly understood. Our investigators and others1,17,18 have also found that AUD individuals show deficits in emotion regulation, including affective flattening and mesocorticolimbic hypo-responsiveness to emotional stimuli (i.e. dampened emotional reactivity), which contribute to increased craving and relapse.19,20 An emerging literature links REM sleep with neural emotion regulation.21-26 A proposed homeostatic model26 posits that the unique neurobiology of REM sleep (e.g. cholinergic activation combined with adrenergic suppression) enables it to fulfil a role in promoting next-day adaptive emotional functioning. In support of this model, selective reduction of REM sleep in healthy subjects leads to next-day hyper-reactivity to emotional stimuli.27 Also, lower prefrontal gamma (30-40Hz) electroencephalographic (EEG) activity in REM sleep, indexing reduced central adrenergic activity,28 relates to more appropriate reactivity to emotions, reduced amygdala (AMYG) activity, and increased ventromedial prefrontal cortex (vmPFC) functional connectivity the next day.29 Our pilot data suggest that central adrenergic activity is not effectively suppressed in REM sleep (higher frontal gamma), suggesting that the REM sleep-emotion recalibration link is dysfunctional in AUD. Our investigators, therefore, propose here that REM sleep in recovering AUD patients is maladaptive for emotion regulation, affecting neural pathways of emotion processing, contributing to their poor emotion regulation. Baker and Műller-Oehring combine their expertise in AUD, sleep, fMRI, and emotion research, and use a novel experimental within-subject, counterbalanced design with two experimental conditions: uninterrupted sleep and selective REM sleep reduction, each followed by a morning fMRI scan. Our investigators will investigate relationships between PSG-measures of sleep, specifically REM sleep, and next-day emotional functioning (fMRI) in AUD patients, abstinent for \> 6 months, compared to age- and sex-matched CTL. Our investigators will also evaluate whether the extent to which AUD individuals demonstrate a capacity to normalize neural emotional regulation after a REM sleep reduction challenge relates to normalization of behavioral measures of emotion reactivity, regulatory control ability, and levels of craving. Aim 1: Unterinterrupted sleep: Determine relationships between sleep and next-day neural emotional functioning in AUD. Hypothesis (H)1: (a) Relative to CTL, AUD will have poorer sleep quality (shorter sleep duration, more awakenings) and exaggerated REM sleep features (excess duration and intensity \[number of eye movements\]) and (b) blunted next-day negative emotional tone defined as reduced neural reactivity to emotional content (blunted mesolimbic responsiveness and less functional connectivity among emotion networks \[amygdala-PFC\]), H2: Higher REM sleep prefrontal gamma (\>30Hz) EEG activity will predict impaired next-day mPFC engagement and top-down control of the amygdala in AUD and CTL. Aim 2: REM sleep disruption: Determine direct effects of experimental REM sleep reduction on next-day neural emotional functioning in AUD. H3: In AUD and CTL, specific REM sleep reduction will heighten next-day emotion reactivity and amygdala-PFC functional connectivity relative to uninterrupted sleep. H4: After REM sleep reduction, next-day emotional responsiveness and amygdala-PFC functional connectivity in ALC will be similar to that of CTL after uninterrupted sleep. H5: In AUD, greater responsiveness of the emotion system to REM sleep manipulation (i.e. greater increase in emotion reactivity relative to uninterrupted sleep condition), will be associated with better behavioral measures of emotion reactivity and regulation on fMRI tasks, and decreased craving. Together, these studies can advance understanding of mechanistic links between sleep processes and emotional brain function in AUD. Findings will inform an R01 application aimed to investigate effects of sleep manipulation as a pathway to resetting neural emotion regulation circuits for relapse prevention. RESEARCH STRATEGY 2. Significance Alcohol use disorder (AUD) is a complex and multideterministic chronic relapsing disorder suffered by 9.8 million men and 5.3 million women in the US.4 Globally, AUD is the most prevalent of all substance use disorders, associated with substantial costs to the health care system and society due to loss of productivity, injury, early retirement, and mortality.30 In 2010, the cost of excessive alcohol use in the US was $249 billion. 31 Remission rates are low, with fewer than 50% of people with an AUD achieving remission after several years follow-up.32 Breaking the cycle of relapse is a critical goal of current pharmacological and other treatment approaches but with limited success.5 Sleep disruption is ubiquitous in AUD patients and persists for years into sobriety; it is associated with a poor quality of life, psychosocial problems, suicidal ideation, and risk of relapse.33,34 It is unknown how sleep disruption increases relapse risk and existing therapies that target insomnia do not necessarily reduce relapse, although they improve functioning.33 In fact, the functional significance of many aspects of disturbed sleep in AUD remains unclear. There is a need to uncover psycho-physiological processes that link sleep disturbance and relapse, which could be targets for AUD treatment. The significance of this proposal lies in the combined investigation of sleep physiology and neural emotion regulation pathways to determine whether the REM sleep-emotion regulation link is dysfunctional in AUD. Findings will inform a future R01 proposal to harness the power of sleep to restore emotional balance, reduce relapse, and lead to better management of AUD. Focus on REM sleep in abstinent AUD. Poor sleep is common in AUD during periods of drinking, acute withdrawal, subacute, and chronic abstinence.13 Sleep disturbances can persist for years into sobriety, with long sleep onset latency, sleep fragmentation, and prominent abnormalities in polysomnographic (PSG) measures of slow wave sleep (SWS) and REM sleep.34 Two longitudinal studies showed deficits in SWS and excess REM sleep early in abstinence, persisting into sobriety,13,35 with excess REM sleep even at 27 months.13 Our investigators also showed excess REM sleep (pilot data) in AUD studied up to 2 years into sobriety relative to matched controls,2 supporting others.36-38 REM sleep differences were unrelated to number of days abstinent, suggesting that this sleep alteration is not a rebound effect due to alcohol withdrawal, and may reflect alcohol-induced alterations in neurobiological sleep processes.10 Excess REM sleep, therefore, likely reflects an unhealthy state that is maladaptive. Increased REM sleep pressure (more REM density, shorter latency to REM sleep and/or more REM sleep) predicts relapse (reviewed in 39): Increased "REM pressure" at admission predicted relapse in 80% of AUD at 3-month follow-up.16 Similarly, others found increased REM pressure in early abstinence predicted relapse 6 months later, with 70% correct classification.38 Thus, REM sleep abnormalities (increased REM sleep and REM density) conceptualized as greater REM pressure, are evident long into sobriety in AUD and predict relapse; however, functional significance of REM sleep abnormalities in AUD are poorly understood. Our investigators hypothesize that ineffective REM sleep-related emotional regulation is one pathway to craving and relapse. Healthy REM sleep and emotion regulation. Sleep is key for emotional well-being and plays an adaptive role in the processing of daily emotions (reviewed in 40, 41). REM sleep in particular has a crucial role in emotion modulation.21-26 During REM sleep, there is selective activation of emotion-related brain regions that are involved in emotional processing during wakefulness (e.g. amygdala).42,43 According to the REM sleep emotional homeostasis hypothesis, the unique neurobiological state of REM sleep enables it to promote adaptive emotional next-day functioning.21 Emotion can be decoupled from memory during REM sleep44-47 and REM sleep serves an "emotion recalibration" function, priming key brain regions before an emotional experience to ensure appropriate reaction.21 There is support for these two hypothesized roles of REM sleep at behavioral and neural levels.22,23,27,29,48 After a night of undisturbed sleep, subjective emotional reactivity and amygdala reactivity to previously-encountered (pre-sleep) emotional stimuli were decreased and connectivity between amygdala and ventromedial prefrontral cortex increased.29 In contrast, amygdala reactivity to emotional stimuli increased across a day of wakefulness compared to first encounter (morning). Importantly, participants with the lowest levels of REM sleep high-frequency prefrontal gamma EEG activity (a validated indirect marker of central adrenergic activity28,49-51) showed the largest overnight decrease in emotion reactivity. This effect was specific for gamma frequency, frontal sites, and REM sleep.29 Also, selective REM sleep reduction amplifies behavioral and neural emotional reactivity to visual stimuli in healthy subjects.27 These findings demonstrate the importance of healthy REM sleep physiology in regulating next-day emotions. Most studies have been done in healthy individuals with intact REM sleep physiology. Here, our investigators build on current theory supporting a causal effect of healthy REM sleep on emotion regulation to investigate relationships in AUD. REM sleep physiology is altered in abstinent AUD but whether it contributes to maladaptive emotion regulation (i.e. has functional consequences) is unknown, and is the focus of this proposal. Emotion regulation in AUD. Emotion plays a role in motivating behavior. Alcohol has a high reinforcing value, can change mood and temporarily alleviate feelings of anxiety. Excessive chronic drinking is accompanied by many emotional changes ranging from apathy and emotional flatness to deficits in comprehending emotional information.18,52 According to Koob's model of addiction,53 repeated cycles of drinking and withdrawal result in an allostatic down-regulation of brain systems maintaining emotional tone that then leads to a cycle of negative emotional state and increased relief-motivated drinking. These negative emotional states persist in abstinence.54 Accordingly, deficits in emotion processing are frequently observed in AUD55-61 accompanied by specific abnormalities in the structure and function of neural networks that regulate emotion.62,63 Neuroanatomically, the reinforcing and rewarding alcohol effects are transmitted via dopaminergic and glutamatergic projections in the mesolimbic corticostriatal system, which originates in midbrain nodes, and connects to the limbic system via the nucleus accumbens, amygdala, and hippocampus, and to the medial prefrontal cortex.64,65 Deficits in prefrontal regions affect the higher-order (top-down) processing of emotional states. AUD (relative to CTL) show less anterior cingulate activation during decoding of negative emotional facial expressions.60 Our investigators observed low prefrontal activation to positive emotional content in AUD (relative to CTL) adding to the model of downregulation of the frontal monitoring system in AUD, even for positive-emotion. Thus, prefrontal regions play a role in the regulatory control of emotion-related behavior66 and their involvement is inversely related to amygdala activity.67 It is plausible that the emotional impairments in AUD are due to a dysfunctional prefrontal-amygdala circuitry - the same circuitry involved in REM sleep emotional homeostasis. Maladaptive REM sleep in AUD (excess REM sleep; exaggerated REM features; high REM gamma prefrontal EEG activity) (pilot data) may contribute to imbalanced emotional homeostasis, associated with impaired prefrontal-amygdala regulation. Emotion and reward systems closely interact.68. Incentive stimuli elicit approach behavior. Alcohol cues gain incentive salience with the development of AUD (incentive salience theory69) and can initiate drinking behaviors and decrease responsiveness to negative emotion.70 Neuroadaptations (upregulation of ventral striatal reward and downregulation of limbic emotion pathways18) that occur with chronic alcohol consumption53,71 are crucial for the negative emotional tone in AUD, contributing to increased craving and relapse.1,17,18,63,72-75 Sex differences. Men and women process emotions differently.76,77 AUD women are 2-3 times more likely to have mood disorders than men, and AUD men are twice as likely to have antisocial personality disorders.59 Both male and female AUD have excess REM sleep compared to controls (Table 1).2 It is unknown if the REM sleep-emotional pathway functions differently in men and women in healthy or AUD subjects: our investigators will explore that possibility, which can lead to insights into what may differentially motivate men and women to abuse alcohol.78-80 In summary, our investigators build on literature showing (1) excess REM sleep and (2) deficits in emotion processing in AUD, both linked to relapse. Our investigators explore if maldaptive REM sleep leads to degraded sleep-dependent emotion regulation, contributing to emotional imbalance in AUD individuals vulnerable to emotional regulation difficulties. 3\. INNOVATION There is both theoretical and methodological innovation. Innovation lies in: (1) assessment of mechanisms linking REM sleep features and emotional dysregulation in AUD. Results could advance research beyond singular mechanistic pathways towards multi-level integrated pathways, potentially leading to novel interventions to transform AUD from a recurring relapsing disorder into a treatable, manageable disorder through sleep-emotion modifications; (2) novel use of experimental REM sleep reduction protocol to directly investigate the potential role of REM sleep in emotional dysregulation in AUD. 4\. APPROACH 4.1. Preliminary studies. Dual sleep and MRI studies in AUD. Investigators (Colrain, Baker) have examined PSG sleep and brain structure in AUD and comparison CTL.2,81,82 PSG data are shown in Table 1. AUD had more awakenings, Stage 1, less SWS (particularly male AUD), and more REM (male and female). REM sleep did not correlate with time since last drink, suggesting that REM sleep excess is not a rebound effect after removal of alcohol. Figure 1 shows frontal EEG power in REM sleep. High-frequency gamma power was higher in AUD than CTL. Our investigators hypothesize that higher REM gamma (indirect marker of adrenergic activity associated with less effective REM sleep emotional regulation29) will be associated with impaired mPFC engagement and top-down control of the amygdala in AUD. REM sleep predicts relapse. In another sample of AUD (n=21, 9 female83,84) studied with multiple PSG across 6 months of recovery, REM sleep tended to be higher in those who later relapsed, supporting the literature.16,38 These sleep data show 1) abnormal REM sleep in abstinent AUD (excess amount and high REM sleep frontal gamma power) and 2) our experience in recruiting and studying PSG in AUD over multiple visits. Emotion regulation in AUD.1 Using fMRI, MPI-Müller-Oehring found evidence for a dysfunctional prefrontal emotion regulation system in abstinent AUD (n=26). Figure 2 (upper panel) shows no midbrain-limbic response to neutral color-word Stroop interference1 in AUD and CTL, but a significantly enhanced response to alcohol and emotion words in AUD (but not CTL). The opposite pattern occurred for prefrontal activation (lower panel): CTL activated prefrontal control regions to overcome interference from emotion and alcohol word content, but AUD showed a blunted dlPFC response. This downregulation of the frontal monitoring system is key to the emotion regulation deficit in AUD. Our investigators will use this fMRI task to test whether emotion dysregulation in AUD is linked to dysfunctional REM sleep and particularly enhanced prefrontal gamma EEG activity. Craving: Frontal hypoactivation together with alcohol-sensitized midbrain-limbic hyperactivation, that overlapped with that to negative emotion, represent a neural mechanism through which emotional dysregulation can lead to alcohol craving. Accordingly, in AUD, higher alcohol craving correlated with higher alcohol-related mesolimbic activity (r=.39, p=.025) and blunted emotion-related prefrontal (dlPFC) activity (r=.45, pFDR-corrected=.011).1 Emotion reactivity in AUD.17 Figure 3, upper panel: Viewing alcohol and emotion pictures during fMRI activated mesolimbic-corticostriatal regions, in particular the parahippocampal gyus, amygdala, thalamus, dorsal striatum, insula, and occipital regions and fusiform face area (FFA). Activation of thalamostriatal reward regions overlapped for emotion and alcohol pictures, providing a neurofunctional link for processing of emotion and reward. Lower panel: Compared to CTL, AUD showed blunted insula and PFC reactivity to emotion although they clearly processed emotional faces (e.g. activated fusiform face area, parahippocampal, and lingual regions, even more than in CTL). Craving: In AUD, craving correlated with greater striatal activity to alcohol pictures (Rho=.71, p=.048); also see 85. These task-activated fMRI pilot data show 1) the overlapping alcohol-emotion brain systems in AUD, 2) poor frontal top-down emotion regulation and altered mesolimbic emotion reactivity are potential pathways to craving and relapse.